Human memory can be separated into recognition ("Is that my car?") versus recall ("Where did I park my car?"). The study of recognition memory, such as in identifying individuals in a police lineup, is often aided by the collection of confidence ratings ("I'm 95 percent sure that's the guy who stole my car"). In contrast, recall is traditionally viewed as a highly accurate all or none process (i.e., creating a criminal sketch from scratch). However, recent experiments demonstrate that recall is often fallible, and is perhaps better viewed as lying along a graded continuum similar to recognition. In the proposed studies, we seek to investigate this graded nature of recall by collecting not only recall responses, but also various kinds of confidence ratings. In some cases we will collect these confidence ratings after learning, but before recall (prospective confidence), while in other cases we will collect these confidence ratings following recall responses (retrospective confidence). We have developed a mathematical tool based on signal detection theory that allows us to use these confidence ratings to determine 1) the graded nature of the memory strength that underlies recall;2) issues related to use of the confidence scale (e.g., some people are conservative while others are liberal in their certainty);and 3) the extent to which confidence is based on different factors than actual memory strength. This work is important not only because it will foster a fuller understanding of recall, but additionally because it will determine under what circumstances confidence predicts or indicates accurate recall. In terms of prospectively predicting recall, there are important implications for educational settings involving self-paced learning. In terms of retrospectively indicating recall, there are important implications for legal settings and other situations involving eyewitness testimony. Finally, the techniques and results will be important not only for assessing memory in normal adults, but also in terms of patient populations and drug treatments that have been demonstrated to dissociate recall performance from other forms of memory. Only by collecting confidence ratings and applying the mathematical tools that we are developing can it be determined whether these dissociations reflect real differences in the underlying memory strength, or whether they instead reflect impairments in confidence and memory access. PUBLIC HEALTH RELEVANCE: Memory disorders are traditionally measured through the inability to recall previously studied items but these deficits may reflect a change in the certainty associated with potential responses rather than a deficit in the underlying memory strength. By collecting confidence judgments as well as recall responses, the proposed work will develop a set of measurement tools that can differentiate between the possible sources of a recall deficit. These measurement tools will be relevant to differential diagnoses of memory disorders as well as clinical therapies designed to alleviate memory problems.